Lichen

Annotated lib/sre_constants.py

6:f551873980e5
2016-08-30 Paul Boddie file changeset files shortlog Added PythonLight alternative libraries.
paul@6 1 
#
paul@6 2 
# Secret Labs' Regular Expression Engine
paul@6 3 
#
paul@6 4 
# various symbols used by the regular expression engine.
paul@6 5 
#
paul@6 6 
# Copyright (c) 1998-2001 by Secret Labs AB.  All rights reserved.
paul@6 7 
#
paul@6 8 
# See the sre.py file for information on usage and redistribution.
paul@6 9 
#
paul@6 10 
 
 paul@6     11 
"""Internal support module for sre"""
 
 paul@6     12 
 
 paul@6     13 
# update when constants are added or removed
 
 paul@6     14 
 
 paul@6     15 
MAGIC = 20031017
 
 paul@6     16 
 
 paul@6     17 
# max code word in this release
 
 paul@6     18 
 
 paul@6     19 
MAXREPEAT = 65535
 
 paul@6     20 
 
 paul@6     21 
# SRE standard exception (access as sre.error)
 
 paul@6     22 
# should this really be here?
 
 paul@6     23 
 
 paul@6     24 
class error(Exception):
 
 paul@6     25 
    pass
 
 paul@6     26 
 
 paul@6     27 
# operators
 
 paul@6     28 
 
 paul@6     29 
FAILURE = "failure"
 
 paul@6     30 
SUCCESS = "success"
 
 paul@6     31 
 
 paul@6     32 
ANY = "any"
 
 paul@6     33 
ANY_ALL = "any_all"
 
 paul@6     34 
ASSERT = "assert"
 
 paul@6     35 
ASSERT_NOT = "assert_not"
 
 paul@6     36 
AT = "at"
 
 paul@6     37 
BIGCHARSET = "bigcharset"
 
 paul@6     38 
BRANCH = "branch"
 
 paul@6     39 
CALL = "call"
 
 paul@6     40 
CATEGORY = "category"
 
 paul@6     41 
CHARSET = "charset"
 
 paul@6     42 
GROUPREF = "groupref"
 
 paul@6     43 
GROUPREF_IGNORE = "groupref_ignore"
 
 paul@6     44 
GROUPREF_EXISTS = "groupref_exists"
 
 paul@6     45 
IN = "in"
 
 paul@6     46 
IN_IGNORE = "in_ignore"
 
 paul@6     47 
INFO = "info"
 
 paul@6     48 
JUMP = "jump"
 
 paul@6     49 
LITERAL = "literal"
 
 paul@6     50 
LITERAL_IGNORE = "literal_ignore"
 
 paul@6     51 
MARK = "mark"
 
 paul@6     52 
MAX_REPEAT = "max_repeat"
 
 paul@6     53 
MAX_UNTIL = "max_until"
 
 paul@6     54 
MIN_REPEAT = "min_repeat"
 
 paul@6     55 
MIN_UNTIL = "min_until"
 
 paul@6     56 
NEGATE = "negate"
 
 paul@6     57 
NOT_LITERAL = "not_literal"
 
 paul@6     58 
NOT_LITERAL_IGNORE = "not_literal_ignore"
 
 paul@6     59 
RANGE = "range"
 
 paul@6     60 
REPEAT = "repeat"
 
 paul@6     61 
REPEAT_ONE = "repeat_one"
 
 paul@6     62 
SUBPATTERN = "subpattern"
 
 paul@6     63 
MIN_REPEAT_ONE = "min_repeat_one"
 
 paul@6     64 
 
 paul@6     65 
# positions
 
 paul@6     66 
AT_BEGINNING = "at_beginning"
 
 paul@6     67 
AT_BEGINNING_LINE = "at_beginning_line"
 
 paul@6     68 
AT_BEGINNING_STRING = "at_beginning_string"
 
 paul@6     69 
AT_BOUNDARY = "at_boundary"
 
 paul@6     70 
AT_NON_BOUNDARY = "at_non_boundary"
 
 paul@6     71 
AT_END = "at_end"
 
 paul@6     72 
AT_END_LINE = "at_end_line"
 
 paul@6     73 
AT_END_STRING = "at_end_string"
 
 paul@6     74 
AT_LOC_BOUNDARY = "at_loc_boundary"
 
 paul@6     75 
AT_LOC_NON_BOUNDARY = "at_loc_non_boundary"
 
 paul@6     76 
AT_UNI_BOUNDARY = "at_uni_boundary"
 
 paul@6     77 
AT_UNI_NON_BOUNDARY = "at_uni_non_boundary"
 
 paul@6     78 
 
 paul@6     79 
# categories
 
 paul@6     80 
CATEGORY_DIGIT = "category_digit"
 
 paul@6     81 
CATEGORY_NOT_DIGIT = "category_not_digit"
 
 paul@6     82 
CATEGORY_SPACE = "category_space"
 
 paul@6     83 
CATEGORY_NOT_SPACE = "category_not_space"
 
 paul@6     84 
CATEGORY_WORD = "category_word"
 
 paul@6     85 
CATEGORY_NOT_WORD = "category_not_word"
 
 paul@6     86 
CATEGORY_LINEBREAK = "category_linebreak"
 
 paul@6     87 
CATEGORY_NOT_LINEBREAK = "category_not_linebreak"
 
 paul@6     88 
CATEGORY_LOC_WORD = "category_loc_word"
 
 paul@6     89 
CATEGORY_LOC_NOT_WORD = "category_loc_not_word"
 
 paul@6     90 
CATEGORY_UNI_DIGIT = "category_uni_digit"
 
 paul@6     91 
CATEGORY_UNI_NOT_DIGIT = "category_uni_not_digit"
 
 paul@6     92 
CATEGORY_UNI_SPACE = "category_uni_space"
 
 paul@6     93 
CATEGORY_UNI_NOT_SPACE = "category_uni_not_space"
 
 paul@6     94 
CATEGORY_UNI_WORD = "category_uni_word"
 
 paul@6     95 
CATEGORY_UNI_NOT_WORD = "category_uni_not_word"
 
 paul@6     96 
CATEGORY_UNI_LINEBREAK = "category_uni_linebreak"
 
 paul@6     97 
CATEGORY_UNI_NOT_LINEBREAK = "category_uni_not_linebreak"
 
 paul@6     98 
 
 paul@6     99 
OPCODES = [
 
 paul@6    100 
 
 paul@6    101 
    # failure=0 success=1 (just because it looks better that way :-)
 
 paul@6    102 
    FAILURE, SUCCESS,
 
 paul@6    103 
 
 paul@6    104 
    ANY, ANY_ALL,
 
 paul@6    105 
    ASSERT, ASSERT_NOT,
 
 paul@6    106 
    AT,
 
 paul@6    107 
    BRANCH,
 
 paul@6    108 
    CALL,
 
 paul@6    109 
    CATEGORY,
 
 paul@6    110 
    CHARSET, BIGCHARSET,
 
 paul@6    111 
    GROUPREF, GROUPREF_EXISTS, GROUPREF_IGNORE,
 
 paul@6    112 
    IN, IN_IGNORE,
 
 paul@6    113 
    INFO,
 
 paul@6    114 
    JUMP,
 
 paul@6    115 
    LITERAL, LITERAL_IGNORE,
 
 paul@6    116 
    MARK,
 
 paul@6    117 
    MAX_UNTIL,
 
 paul@6    118 
    MIN_UNTIL,
 
 paul@6    119 
    NOT_LITERAL, NOT_LITERAL_IGNORE,
 
 paul@6    120 
    NEGATE,
 
 paul@6    121 
    RANGE,
 
 paul@6    122 
    REPEAT,
 
 paul@6    123 
    REPEAT_ONE,
 
 paul@6    124 
    SUBPATTERN,
 
 paul@6    125 
    MIN_REPEAT_ONE
 
 paul@6    126 
 
 paul@6    127 
]
 
 paul@6    128 
 
 paul@6    129 
ATCODES = [
 
 paul@6    130 
    AT_BEGINNING, AT_BEGINNING_LINE, AT_BEGINNING_STRING, AT_BOUNDARY,
 
 paul@6    131 
    AT_NON_BOUNDARY, AT_END, AT_END_LINE, AT_END_STRING,
 
 paul@6    132 
    AT_LOC_BOUNDARY, AT_LOC_NON_BOUNDARY, AT_UNI_BOUNDARY,
 
 paul@6    133 
    AT_UNI_NON_BOUNDARY
 
 paul@6    134 
]
 
 paul@6    135 
 
 paul@6    136 
CHCODES = [
 
 paul@6    137 
    CATEGORY_DIGIT, CATEGORY_NOT_DIGIT, CATEGORY_SPACE,
 
 paul@6    138 
    CATEGORY_NOT_SPACE, CATEGORY_WORD, CATEGORY_NOT_WORD,
 
 paul@6    139 
    CATEGORY_LINEBREAK, CATEGORY_NOT_LINEBREAK, CATEGORY_LOC_WORD,
 
 paul@6    140 
    CATEGORY_LOC_NOT_WORD, CATEGORY_UNI_DIGIT, CATEGORY_UNI_NOT_DIGIT,
 
 paul@6    141 
    CATEGORY_UNI_SPACE, CATEGORY_UNI_NOT_SPACE, CATEGORY_UNI_WORD,
 
 paul@6    142 
    CATEGORY_UNI_NOT_WORD, CATEGORY_UNI_LINEBREAK,
 
 paul@6    143 
    CATEGORY_UNI_NOT_LINEBREAK
 
 paul@6    144 
]
 
 paul@6    145 
 
 paul@6    146 
def makedict(list):
 
 paul@6    147 
    d = {}
 
 paul@6    148 
    i = 0
 
 paul@6    149 
    for item in list:
 
 paul@6    150 
        d[item] = i
 
 paul@6    151 
        i = i + 1
 
 paul@6    152 
    return d
 
 paul@6    153 
 
 paul@6    154 
OPCODES = makedict(OPCODES)
 
 paul@6    155 
ATCODES = makedict(ATCODES)
 
 paul@6    156 
CHCODES = makedict(CHCODES)
 
 paul@6    157 
 
 paul@6    158 
# replacement operations for "ignore case" mode
 
 paul@6    159 
OP_IGNORE = {
 
 paul@6    160 
    GROUPREF: GROUPREF_IGNORE,
 
 paul@6    161 
    IN: IN_IGNORE,
 
 paul@6    162 
    LITERAL: LITERAL_IGNORE,
 
 paul@6    163 
    NOT_LITERAL: NOT_LITERAL_IGNORE
 
 paul@6    164 
}
 
 paul@6    165 
 
 paul@6    166 
AT_MULTILINE = {
 
 paul@6    167 
    AT_BEGINNING: AT_BEGINNING_LINE,
 
 paul@6    168 
    AT_END: AT_END_LINE
 
 paul@6    169 
}
 
 paul@6    170 
 
 paul@6    171 
AT_LOCALE = {
 
 paul@6    172 
    AT_BOUNDARY: AT_LOC_BOUNDARY,
 
 paul@6    173 
    AT_NON_BOUNDARY: AT_LOC_NON_BOUNDARY
 
 paul@6    174 
}
 
 paul@6    175 
 
 paul@6    176 
AT_UNICODE = {
 
 paul@6    177 
    AT_BOUNDARY: AT_UNI_BOUNDARY,
 
 paul@6    178 
    AT_NON_BOUNDARY: AT_UNI_NON_BOUNDARY
 
 paul@6    179 
}
 
 paul@6    180 
 
 paul@6    181 
CH_LOCALE = {
 
 paul@6    182 
    CATEGORY_DIGIT: CATEGORY_DIGIT,
 
 paul@6    183 
    CATEGORY_NOT_DIGIT: CATEGORY_NOT_DIGIT,
 
 paul@6    184 
    CATEGORY_SPACE: CATEGORY_SPACE,
 
 paul@6    185 
    CATEGORY_NOT_SPACE: CATEGORY_NOT_SPACE,
 
 paul@6    186 
    CATEGORY_WORD: CATEGORY_LOC_WORD,
 
 paul@6    187 
    CATEGORY_NOT_WORD: CATEGORY_LOC_NOT_WORD,
 
 paul@6    188 
    CATEGORY_LINEBREAK: CATEGORY_LINEBREAK,
 
 paul@6    189 
    CATEGORY_NOT_LINEBREAK: CATEGORY_NOT_LINEBREAK
 
 paul@6    190 
}
 
 paul@6    191 
 
 paul@6    192 
CH_UNICODE = {
 
 paul@6    193 
    CATEGORY_DIGIT: CATEGORY_UNI_DIGIT,
 
 paul@6    194 
    CATEGORY_NOT_DIGIT: CATEGORY_UNI_NOT_DIGIT,
 
 paul@6    195 
    CATEGORY_SPACE: CATEGORY_UNI_SPACE,
 
 paul@6    196 
    CATEGORY_NOT_SPACE: CATEGORY_UNI_NOT_SPACE,
 
 paul@6    197 
    CATEGORY_WORD: CATEGORY_UNI_WORD,
 
 paul@6    198 
    CATEGORY_NOT_WORD: CATEGORY_UNI_NOT_WORD,
 
 paul@6    199 
    CATEGORY_LINEBREAK: CATEGORY_UNI_LINEBREAK,
 
 paul@6    200 
    CATEGORY_NOT_LINEBREAK: CATEGORY_UNI_NOT_LINEBREAK
 
 paul@6    201 
}
 
 paul@6    202 
 
 paul@6    203 
# flags
 
 paul@6    204 
SRE_FLAG_TEMPLATE = 1 # template mode (disable backtracking)
 
 paul@6    205 
SRE_FLAG_IGNORECASE = 2 # case insensitive
 
 paul@6    206 
SRE_FLAG_LOCALE = 4 # honour system locale
 
 paul@6    207 
SRE_FLAG_MULTILINE = 8 # treat target as multiline string
 
 paul@6    208 
SRE_FLAG_DOTALL = 16 # treat target as a single string
 
 paul@6    209 
SRE_FLAG_UNICODE = 32 # use unicode locale
 
 paul@6    210 
SRE_FLAG_VERBOSE = 64 # ignore whitespace and comments
 
 paul@6    211 
SRE_FLAG_DEBUG = 128 # debugging
 
 paul@6    212 
 
 paul@6    213 
# flags for INFO primitive
 
 paul@6    214 
SRE_INFO_PREFIX = 1 # has prefix
 
 paul@6    215 
SRE_INFO_LITERAL = 2 # entire pattern is literal (given by prefix)
 
 paul@6    216 
SRE_INFO_CHARSET = 4 # pattern starts with character from given set